Apoptotic Genes As Potential Markers of Metastatic Phenotype in Human Osteosarcoma Cell Lines

17-31 10/12/07 14:53 Page 17

INTERNATIONAL JOURNAL OF ONCOLOGY 32: 17-31, 2008 17

Apoptotic genes as potential markers of metastatic phenotype in human osteosarcoma cell lines

CINZIA ZUCCHINI1, ANNA ROCCHI2, MARIA CRISTINA MANARA2, PAOLA DE SANCTIS1, CRISTINA CAPANNI3, MICHELE BIANCHINI1, PAOLO CARINCI1, KATIA SCOTLANDI2 and LUISA VALVASSORI1

1Dipartimento di Istologia, Embriologia e Biologia Applicata, Università di Bologna, Via Belmeloro 8, 40126 Bologna; 2Laboratorio di Ricerca Oncologica, Istituti Ortopedici Rizzoli; 3IGM-CNR, Unit of Bologna, c/o Istituti Ortopedici Rizzoli, Via di Barbiano 1/10, 40136 Bologna, Italy

Received May 29, 2007; Accepted July 19, 2007

Abstract. Metastasis is the most frequent cause of death among malignant primitive bone tumor, usually developing in children patients with osteosarcoma. We have previously demonstrated and adolescents, with a high tendency to metastasize (2). in independent experiments that the forced expression of Metastases in osteosarcoma patients spread through peripheral L/B/K ALP and CD99 in U-2 OS osteosarcoma cell lines blood very early and colonize primarily the lung, and later markedly reduces the metastatic ability of these cancer cells. other skeleton districts (3). Since disseminated hidden micro- This behavior makes these cell lines a useful model to assess metastases are present in 80-90% of OS patients at the time the intersection of multiple and independent gene expression of diagnosis, the identification of markers of invasiveness signatures concerning the biological problem of dissemination. and metastasis forms a target of paramount importance in With the aim to characterize a common transcriptional profile planning the treatment of osteosarcoma lesions and enhancing reflecting the essential features of metastatic behavior, we the prognosis. employed cDNA microarrays to compare the gene expression In this study we performed a statistical analysis of multiple profiles of L/B/K ALP- and CD99-transfected osteosarcoma gene expression datasets concerning the common biological clones showing low metastatic ability with those of osteo- problem of metastasis. Our comprehensive analysis includes sarcoma cell lines showing contrasting behavior. Changes the gene expression datasets derived from two experimental in gene expression were validated by real-time PCR and models previously obtained from the U-2 OS osteosarcoma immunohistochemistry in independent samples. In our study we cell lines characterized by a common reversal of malignancy identified several differentially expressed genes (GADD45α, and lack of metastatic phenotype. In particular, we have VCP, DHX9, survivin, α-catulin, ARPC1B) related to growth demonstrated in independent experiments that the forced arrest and apoptosis. Most of these genes are functionally expression of L/B/K ALP and likewise CD99 in U-2 OS related with the nuclear factor (NF)-κB cell survival pathway osteosarcoma cell lines markedly reduces the invasiveness that appeared to be inhibited in the less malignant osteosarcoma and metastatic ability of these cancer cells (4,5). Alkaline cells. Hence, we propose the inhibition of the NF-κB pathway phosphatases (ALPs) are a family of cell surface glycoproteins as a rational strategy for effective management of human that catalyze the hydrolysis of phospho-monoesters and release osteosarcoma. inorganic phosphate. Liver-bone-kidney (L/B/K) ALP, one of the four major isoenzymes belonging to this family, participates Introduction in bone mineralization. The precise biochemical function of bone ALP activity is still unknown, but in U-2 OS osteo- Metastasis is the dissemination of cancer cells from the primary sarcoma cell lines transfected with the full-length L/B/K ALP tumor to a distant organ and represents the most frequent cause gene a reduction in tumorigenic and metastatic ability has of death for patients with cancer (1). Osteosarcoma (OS) is a been observed associated with high levels of cellular activity of L/B/K ALP (4). CD99 is a transmembrane glycoprotein (6) ______encoded by the MIC2 gene (7), which has been implicated in several cellular processes, such as cell adhesion, apoptosis and differentiation (8-12). Manara et al (5) have demonstrated Correspondence to: Dr Katia Scotlandi, Laboratorio di Ricerca that the forced expression of CD99 in two osteosarcoma cell Oncologica, Istituti Ortopedici Rizzoli, Via di Barbiano 1/10, 40136 lines significantly inhibits growth in anchorage independence Bologna, Italy E-mail: [email protected] as well as cell migration and leads to abrogation of tumorigenic and metastatic ability. Key words: gene expression profile, microarray, U2-OS, metastases, The fact that U-2 OS osteosarcoma cell lines transfected nuclear factor-κB with both L/B/K ALP and CD99 show identical behavior (that is, marked reduction of metastatic ability), makes these 17-31 10/12/07 14:53 Page 18

18 ZUCCHINI et al: APOPTOTIC GENES AS PREDICTIVE MARKERS IN OSTEOSARCOMA

Table I. In vitro and in vivo biological features associated with malignancy of U-2 OS transfected clones. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Cell line In vitro parametersa In vivo behaviourb ––––––––––––––––––––––––––––––– –––––––––––––––––––––––––––––––––––––––––––––––– Soft-agar growth Migration Mice with tumor/total Mice with metastases/total ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– U-2 OS 742±102 215±40 7/10 (70%) 16/19 (84%) U-2/Empty 574±68 218±40 3/5 (60%) 9/10 (90%) U-2/ALP23 723±16 183±20 3/5 (60%) 10/10 (100%) U-2/ALP28 109±7c 26±5c 0/5 (0%) 3/7 (43%) U-2/ALP40 118±17c 92±6c 0/5 (0%) 4/14 (29%) U-2/CD99wt57 67±17c 90±18c 0/5 (0%) 0/10 (0%) U-2/CD99wt 136 124±26c 64±17c 0/5 (0%) 7/15 (47%) ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– aSoft agar growth was measured by counting cell colonies 14 days after seeding of 10,000 cells in 0.33% agarose. Migration was assessed by counting the number of cells that migrated towards the filter to reach the lower chamber base of a Trans-well chamber. Cells (105) in IMDM plus 1% FBS were seeded in the upper compartment, whereas IMDM plus 10% FBS was placed in the lower compartment of the chamber as a chemoattractant. bTumorigenicity was determined after subcutaneous injection of 30x106 cells in female 4-5 week old Cr1:nu/nu (CD-1)BR athymic mice (Charles River Italia, Como). Experimental metastatic ability was evaluated after injection of 2x106 cells into a tail lateral vein. For detailed methods see Manara et al (4,5). cP<0.05, Student's t-test. –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

cell lines a useful model to assess the intersection of multiple quality of RNA samples was determined by electrophoresis and independent gene expression signatures concerning the through agarose gels and staining with ethidium bromide; the common biological problem of dissemination. In particular, 18S and 28S RNA bands were visualized under UV light. To in our study we employed cDNA microarray technology to perform microarray hybridization, two independent extractions compare the gene expression profiles of L/B/K ALP- and of RNAs were performed from each cell clone. CD99-transfected osteosarcoma clones showing low metastatic ability with those of osteosarcoma cell lines showing cDNA microarray hybridization. Hybridizations were contrasting behavior, the purpose being to characterize a performed on Human 1 cDNA Microarray slides (Agilent common transcriptional profile reflecting the essential features Technologies, Inc., Palo Alto, CA) containing 16,281 cDNA of metastatic behavior. sequences. Total RNA was used to obtain labeled cDNA, according to the manufacturer's instructions (Agilent Direct- Materials and methods Label cDNA synthesis kit protocol, Agilent Technologies). cDNA from osteosarcoma cell lines transfected with L/B/K Cell lines. The parental osteosarcoma cell line U-2 OS was ALP and with CD99 was labeled with Cyanine 5-dCTP obtained from the American Type Culture Collection (Cy5), while parental U-2/OS cDNA was labeled with Cyanine (Manassas, VA). Cells overexpressing L/B/K ALP or CD99 3-dCTP (Cy3). In brief, 20 μg of total RNA was reverse- were obtained by using the calcium-phosphate transfection transcribed with a specific primer mixture using Moloney method previously characterized (4,5). Transfectants were murine leukemia virus reverse transcriptase (Agilent Direct- selected in 500 μg/ml neomycin (Sigma-Aldrich, St. Louis, Label cDNA synthesis kit) in the presence of Cy5 or Cy3 MO) and maintained in a selective medium for a maximum of (Perkin-Elmer Life Sciences Inc., Boston, MA) at 42˚C for 1 h. eight in vitro passages before in vitro and in vivo characteriz- RNaseI A (Agilent Technologies) was then added to degrade ation. In particular, U-2/ALP28, and U-2/ALP40, showing high the RNA. Labeled cDNAs were purified with the QIAquick levels of L/B/K ALP expression and activity, and U-2/CD99 PCR purification kit (Qiagen, GmbH, Germany). Additional clones (U2-CD99wt57 and U2-CD99wt136), overexpressing washes with 35% (w/v) guanidine hydrocloride were performed CD99 at the cell surface, were chosen for their low level of to ensure efficient removal of the unincorporated dye labeled malignancy. Table I summarizes the previously published nucleotides. Equivalent amounts of Cy5- and Cy3-cDNA (4,5) U-2 OS transfectant features associated with cancer were combined, vacuum dried, re-suspended in deposition malignancy relevant for this study. Cells transfected with hybridization buffer (Agilent Technologies) with human the empty vector pcDNA3 (5) or clone U-2/ALP23, which Cot-1 (Invitrogen) and deposition control targets (Qiagen) shows very low levels of L/B/K ALP activity (4), were used and hybridized to microarray slides for approximately 17 h, as negative controls. Cells were routinely cultured in Iscove's according to the manufacturer's instructions. The slides were modified Dulbecco's medium (IMDM) (Invitrogen, Paisley, washed with 0.5X SSC + 0.01% SDS for 10 min and with Scotland), supplemented with 100 U/ml penicillin, 100 μg/ml 0.06X SSC for 5 min. streptomycin and 10% inactivated fetal bovine serum (FBS) (Cambrex BioScience Verviers, Verviers, Belgium) and Microarray analysis. Hybridized slides were scanned using the maintained at 37˚C in a humidified 5% CO2 atmosphere. GenePix 4000A scanner (Axon Instruments, Foster City, CA) ensuring that the total number of saturated spots (saturation RNA isolation. Total RNA was extracted using the TRIzol fluorescence value: 65,000) was <10. Images were analyzed extraction kit (Invitrogen Life Technologies, Paisley, UK). The using GENEPIX PRO 3.0 software (Axon Instruments). 17-31 10/12/07 14:53 Page 19

INTERNATIONAL JOURNAL OF ONCOLOGY 32: 17-31, 2008 19

GenePix Post-Processing Script (GP3) was used in order was also normalized to U-2 OS mRNA, used as a calibrator -ΔΔCT to filter and normalize raw microarray data from GenePix and was expressed as: 2 , where ΔCT = CT target genes -

image analysis (13) (global normalization; percentile of mean CT GAPDH, and ΔΔCT = ΔCT sample - ΔCT calibrator. All signal across array for normalization factor: 90; threshold samples were resolved in a 2% agarose gel to confirm the value: median background spot intensity + 3 standand deviation PCR specificity. of background). Expression tables were assembled from GP3 output files and genes that proved to be filtered and preserved Analysis of apoptosis. Detection and quantification of apoptotic in 100% of the arrays were analyzed using the Significance cells was obtained by flow cytometric analysis (FACSCalibur, Analysis of Microarray (SAM) program (14). SAM identifies Becton Dickinson) of annexin-V-FITC-labelled cells. This genes with statistically significant changes in expression by test was performed according to the manufacturer's instructions performing a set of gene-specific t-tests. Each gene has a score (Medical and Biological Laboratories, Naka-ku Nagaya, Japan). (d) assigned on the basis of its change in gene expression PI incorporation was evaluated in combination with the relative to the standard deviation of repeated measurements fluorescent signal intensity to allow for discrimination of for that gene. Only genes showing a score of >2.5 or <-2.5 necrotic and apoptotic cells. were considered, and among these, only genes up-regulated or down-regulated in at least 85% of the arrays were selected Poly-HEMA assay. Six-well plates were treated with poly-2 and employed for functional analysis. hydroxyethylmethacrylate (poly-HEMA) (Sigma) following the method of Folkman and Moscona (15). Briefly, wells were Gene ontology (GO). We used FatiGO to assign the 253 treated with a 1-ml solution of poly-HEMA diluted in 95% differently expressed genes to non-mutually exclusive ethanol (12 mg/ml) and left to dry at room temperature. Cells categories regarding biological processes. FatiGO is a web (250,000/well) were seeded in IMDM 10% FBS and incubated interface (http://fatigo.bioinfo.cipf.es) which carries out at 37˚C in a humidified 5% CO2 atmosphere. Viable and dead simple data mining using GO for DNA microarray data. The cells were counted after 24, 48, and 72 h. Detection and data mining consists in assigning the most characteristic GO quantification of apoptotic cells as well as of cell cycle phases term to each cluster. When we searched for the distribution was obtained according to the procedures described above. of genes in a specific ontology (e.g. a biological process) we selected the ontology and also the GO term level (=3). Immunostainings on adherent fixed cells. Cells were seeded in IMDM 10% FBS and grown on coverslips for 48 h before Real-time PCR. Total RNA (1 μg) was denatured at 65˚C for being fixed in 4% paraformaldehyde at room temperature 10 min and then reverse-transcribed in a 100-μl reaction and permeabilized with 0.15% Triton X-100 in PBS. All mixture containing 500 μM of each dNTP, 125 U of Multi- preparations were incubated with PBS containing 4% BSA to Scribe reverse transcriptase (Applied Biosystems, Foster saturate non-specific binding. Immunofluorescence staining City, CA), 40 U of RNase inhibitor (Applied Biosystems), for cyclin D1 and NF-κB was performed with the primary 2.5 μM oligo d(T), 1X TaqMan RT Buffer, and 5 mM MgCl2 cyclin D1 (H-295) MAb (Santa Cruz Biotechnolgy, Inc.) at 48˚C for 40 min. Reactions performed in the absence of (diluted 1:10) or anti NF-κBp50 (H-119) MAb (Santa Cruz enzyme or RNA were used as negative controls. Gene-specific Biotechnolgy, Inc.) (diluted 1:20), respectively. FITC- primers were designed using Primer Express software (Applied conjugated phalloidin (5 U/ml) (Sigma) was applied for 30 h Biosystems): Glyceraldehyde-3-phosphate dehydrogenase at room temperature to stain and visualize actin filaments. (GAPDH) primers and probe as follows: forward 5' GAAGG TGAAGGTCGGAGTC 3', reverse 5' GAAGATGGTGATG Wound-healing assay. Monolayer wounds were made using a GGATTTC 3'; probe: FAM-GAA 5' GCTTCCCGTTCTCA pipette tip on confluent cells. Cell migration was visualized, GCC 3'. Actin Related Protein 2/3 Complex, Subunit 1B, at regular intervals of time, at x100 magnification using an 41 kDa (ARPC1B): forward 5' TCAAGTGTCGGATCTTT inverted microscope (Nikon Diaphot, Melville, NY), and TCAGC 3', reverse 5' AGTTCCCCAAAGGGCATCTT 3'. photographed with a D70s Nikon digital camera (Nikon). α-catulin [Catenin (cadherin-associated protein), α-like1, CTNNAL1]: forward 5' AGTGGGGAAGATGCAGTCC 3', Nuclei extraction and NF-κB Western blotting. The cell pellet reverse 5' CAAATGACAGCTAAACAGCAGC 3'. Survivin was resuspended in a buffer containing 10 mM Tris, pH 7.8, (Baculoviral IAP Repeat-Containing 5, BIRC5): forward 5' 1% NP-40, 10 mM 2-mercaptoethanol and protease inhibitors. TGGCCCAGTGTTTCTTCTGC 3', reverse 5' CAACCGGA Separation of nuclei was obtained by hypotonic shock and CGAATGCTTTTT 3'. Valosin-containing protein (VCP): shearing; nuclei were obtained as pellet by a 300 x g centri- forward 5' GTAACCTGGGAAGACATCGGG 3', reverse 5' fugation at 4˚C. Cytoplasmic fractions were clarified by GGAATTTGTCTGGGTGCTCC 3'. A SYBR-Green PCR centrifugation at 600 g. Each fraction was then resuspended Master mix (Applied Biosystems) was used with 1 ng of in loading buffer, subjected to sodium dodecylsulphate poly- cDNA and with 100-400 nM of the primers. A negative control acrylamide gel electrophoresis (SDS PAGE) and Western without any cDNA template was run with every assay. All PCR blot analysis. The expression of NF-κB in the nuclei was reactions were performed using an ABI PRISM 7900 sequence evaluated using anti-NF-κBp50 (H-119) MAb (Santa Cruz detection system (PE Applied Biosystems). The target gene Biotechnolgy, Inc.) (diluted 1:100). The absence of cytoplasmic mRNA was quantified by measuring the CT (threshold cycle) contamination in the nuclear fraction was checked by staining to determine the relative expression. Data were normalized to the cellular fractions with anti-GAPDH Mab (FL-335) (diluted GAPDH. The relative expression of the different mRNAs 1:2000). 17-31 10/12/07 14:53 Page 20

20 ZUCCHINI et al: APOPTOTIC GENES AS PREDICTIVE MARKERS IN OSTEOSARCOMA

Table II. Up-regulated and down-regulated expressed genes in both L/B/K ALP and CD99 U2-OS transfected clones showing lack of metastatic ability compared with parental U2-OS cell line. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Gene symbol Gene name Unigene Score (d) ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Up-regulated genes KIF23 Kinesin family member 23 Hs.270845 5.872 CRYAB Homo sapiens crystallin, α B Hs.408767 5.718 OXA1L Oxidase (cytochrome c) assembly 1-like Hs.151134 3.853 DUSP12 Dual specificity phosphatase 12 Hs.416216 3.545 VGF VGF nerve growth factor inducible Hs.439672 3.491 GNG11 Guanine nucleotide binding protein (G protein), γ 11 Hs.83381 3.401 MTHFD2 Methylenetetrahydrofolate dehydrogenase (NADP+ dependent) 2, methenyltetrahydrofolate Hs.469030 3.385 cyclohydrolase RAD51C RAD51 homolog C (S. cerevisiae) Hs.412587 3.382 BUB1 Budding uninhibited by benzimidazoles 1 homolog (yeast) Hs.469649 3.38 ANXA1 Annexin A1 Hs.494173 3.376 TAF13 TAF13 RNA polymerase II, TATA box binding protein (TBP)-associated factor, 18 kDa Hs.502508 3.339 PMAIP1 Phorbol-12-myristate-13-acetate-induced protein 1 Hs.96 3.32 SLC38A2 Solute carrier family 38, member 2 Hs.221847 3.207 ANAPC10 Anaphase promoting complex subunit 10 Hs.480876 3.162 SLC22A11 Solute carrier family 22 (organic anion/cation transporter), member 11 Hs.220844 3.139 MTX2 metaxin 2 Hs.470728 3.104 SBDS Shwachman-Bodian-Diamond syndrome Hs.110445 3.08 RCN2 Reticulocalbin 2, EF-hand calcium binding domain Hs.79088 3.079 PBK PDZ binding kinase Hs.104741 3.059 PCDHGC3 protocadherin γ subfamily C, 3 gene, exon 1 Hs.368160 3.023 TJP2 Tight junction protein 2 (zona occludens 2) Hs.50382 3.006 COPS4 COP9 constitutive photomorphogenic homolog subunit 4 (Arabidopsis) Hs.190384 2.974 CTGF Connective tissue growth factor Hs.410037 2.93 TRO Trophinin Hs.434971 2.899 SNAPC1 Small nuclear RNA activating complex, polypeptide 1, 43 kDa Hs.179312 2.89 TPT1 Tumor protein, translationally-controlled 1 Hs.374596 2.887 RPL9 Ribosomal protein L9 Hs.513083 2.883 NAT5 N-acetyltransferase 5 (ARD1 homolog, S. cerevisiae) Hs.368783 2.852 EIF3S6 Eukaryotic translation initiation factor 3, subunit 6 48 kDa Hs.405590 2.848 ADAM9 A disintegrin and metalloproteinase domain 9 (meltrin γ) Hs.591852 2.821 DDEF1 Development and differentiation enhancing factor 1 Hs.106015 2.794 RTN4 Reticulon 4 Hs.645283 2.785 COX6C Cytochrome c oxidase subunit VIc Hs.351875 2.778 PGRMC1 Progesterone receptor membrane component 1 Hs.90061 2.776 GAD2 Glutamate decarboxylase 2 (pancreatic islets and brain, 65 kDa) Hs.231829 2.768 TBCA Tubulin-specific chaperone a Hs.495912 2.766 Transcribed locus Hs.598710 2.742 WSB2 WD repeat and SOCS box-containing 2 Hs.506985 2.726 PRNP prion protein (p27-30) (Creutzfeldt-Jakob disease, Gerstmann-Strausler-Scheinker syndrome, fatal Hs.472010 2.706 familial insomnia) DDX18 DEAD (Asp-Glu-Ala-Asp) box polypeptide 18 Hs.363492 2.702 PHLDA2 Pleckstrin homology-like domain, family A, member 2 Hs.154036 2.702 TM4SF8 Tetraspanin 3 Hs.5062 2.681 HSBP1 Heat shock factor binding protein 1 Hs.250899 2.673 SDHD Succinate dehydrogenase complex, subunit D, integral membrane protein Hs.356270 2.663 RPS15A Ribosomal protein S15a Hs.370504 2.66 NDUFB3 NADH dehydrogenase (ubiquinone) 1 ß subcomplex, 3, 12 kDa Hs.109760 2.649 17-31 10/12/07 14:53 Page 21

INTERNATIONAL JOURNAL OF ONCOLOGY 32: 17-31, 2008 21

Down-regulated genes NOLC1 Nucleolar and Coiled-Body Phosphoprotein 1 Hs.523238 -6.736 TPM2 Tropomyosin 2 (ß) Hs.300772 -6.692 EEF1D Eukaryotic Translation Elongation Factor 1 δ (Guanine Nucleotide Exchange Protein) Hs.333388 -5.508 TPM3 Tropomyosin 3 Hs.129512 -5.457 THOC4 THO Complex 4 Hs.534385 -5.237 HSPCB Heat Shock 90 kDa Protein 1, ß Hs.509736 -5.066 HNRPA2B1 Heterogeneous Nuclear Ribonucleoprotein A2/B1 Hs.487774 -5.018 LIG1 Ligase I, DNA, ATP-Dependent Hs.1770 -4.922 NDUFV1 NADH Dehydrogenase (Ubiquinone) Flavoprotein 1, 51 kDa Hs.7744 -4.908 TUBA1 Tubulin, α 1 (Testis Specific) Hs.75318 -4.838 ECHS1 Enoyl Coenzyme A Hydratase, Short Chain, 1, Mitochondrial Hs.76394 -4.823 HADHA Hydroxyacyl-Coenzyme A Dehydrogenase/3-Ketoacyl-Coenzyme A Thiolase/Enoyl-Coenzyme A Hs.516032 -4.748 Hydratase (Trifunctional Protein), α subunit LMAN2 Lectin, Mannose-Binding 2 Hs.75864 -4.721 TTN Titin Hs.134602 -4.698 GANAB Glucosidase, α; Neutral AB Hs.595071 -4.589 TUBB4Q Tubulin, ß Polypeptide 4, Member Q Hs.351544 -4.58 TSR2 TSR2, 20S rRNA Accumulation, Homolog (S. Cerevisiae) Hs.522662 -4.502 PTMA Prothymosin, α (gene sequence 28) Hs.459927 -4.489 DHX9 DEAH (Asp-Glu-Ala-His) box polypeptide 9 Hs.191518 -4.485 17-31 10/12/07 14:53 Page 22

22 ZUCCHINI et al: APOPTOTIC GENES AS PREDICTIVE MARKERS IN OSTEOSARCOMA

Table II. Continued. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Gene symbol Gene name Unigene Score (d) ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– SMARCA4 SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily a, member 4 Hs.327527 -4.473 TUBAL3 Tubulin, α-like 3 Hs.163079 -4.436 VCP Valosin-Containing Protein Hs.529782 -4.412 SYNGR2 Synaptogyrin 2 Hs.464210 -4.251 SFN Stratifin Hs.523718 -4.25 EIF4H Eukaryotic Translation Initiation Factor 4H Hs.520943 -4.217 PCOLN3 Procollagen (Type III) N-Endopeptidase Hs.589427 -4.216 ETFB Electron-transfer-flavoprotein, ß polypeptide Hs.74047 -4.097 HLA-E Major Histocompatibility Complex, Class I, E Hs.118354 -4.089 STK24 Serine/threonine kinase 24 (STE20 homolog, yeast) Hs.508514 -4.076 SF3B2 Splicing factor 3b, subunit 2, 145 kDa Hs.406423 -4.022 SRF Serum Response Factor (C-Fos Serum Response Element-Binding Transcription Factor) Hs.520140 -4.01 FKBP2 FK506 binding protein 2, 13 kDa Hs.227729 -4.009 ENO2 Enolase 2 (γ, neuronal) Hs.511915 -3.99 CALR Calreticulin Hs.515162 -3.919 HNRPH3 Heterogeneous Nuclear Ribonucleoprotein H3 (2H9) Hs.652156 -3.909 NCL Nucleolin Hs.79110 -3.908 DKC1 Dyskeratosis Congenita 1, Dyskerin Hs.4747 -3.891 TUBA1 Tubulin, α 1 Hs.75318 -3.889 PIK4CB Phosphatidylinositol 4-Kinase, Catalytic, ß polypeptide Hs.632465 -3.871 CDC37 CDC37 Cell Division Cycle 37 Homolog (S. Cerevisiae) Hs.160958 -3.868 AMFR Autocrine Motility Factor Receptor Hs.295137 -3.867 CHST10 Carbohydrate Sulfotransferase 10 Hs.516370 -3.82 UBE2S Ubiquitin-Conjugating Enzyme E2S Hs.396393 -3.775 DYNC1H1 Dynein, Cytoplasmic 1, Heavy Chain 1 Hs.569312 -3.701 PYCR1 Pyrroline-5-Carboxylate Reductase 1 Hs.458332 -3.69 EML4 Echinoderm Microtubule Associated Protein Like 4 Hs.593614 -3.663 SLC25A6 Solute Carrier Family 25 (Mitochondrial Carrier; Adenine Nucleotide Translocator), member 6 Hs.350927 -3.646 IFRD2 Interferon-Related Developmental Regulator 2 Hs.315177 -3.63 FEN1 Flap Structure-Specific Endonuclease 1 Hs.409065 -3.629 CDC20 CDC20 Cell Division Cycle 20 Homolog (S. Cerevisiae) Hs.524947 -3.577 GNB2 Guanine Nucleotide Binding Protein (G Protein), ß polypeptide 2 Hs.185172 -3.564 B4GALT6 UDP-Gal:ßglcnac ß 1,4- Galactosyltransferase, polypeptide 6 Hs.591063 -3.557 PKM2 Pyruvate Kinase, muscle Hs.534770 -3.551 LRPAP1 Low Density Lipoprotein Receptor-Related Protein Associated Protein 1 Hs.533136 -3.547 TK1 Thymidine Kinase 1, Soluble Hs.515122 -3.537 MCAM Melanoma Cell Adhesion Molecule Hs.599039 -3.532 NDUFB7 NADH Dehydrogenase (Ubiquinone) 1 ß subcomplex, 7, 18 kDa Hs.532853 -3.53 ACTG1 Actin, γ 1 Hs.514581 -3.53 U2AF1 U2(RNU2) Small Nuclear RNA Auxiliary Factor 1 Hs.365116 -3.502 PEBP1 Phosphatidylethanolamine Binding Protein 1 Hs.652392 -3.5 RBM4 RNA Binding Motif Protein 4 Hs.533712 -3.49 MLC1SA Myosin, Light Chain 6B, Alkali, smooth muscle and non-muscle Hs.632731 -3.481 DHX38 DEAH (Asp-Glu-Ala-His) Box Polypeptide 38 Hs.570079 -3.472 TUBB2B Tubulin, ß 2B Hs.300701 -3.468 ARF1 ADP-Ribosylation Factor 1 Hs.286221 -3.466 TWF2 Twinfilin, Actin-Binding Protein, Homolog 2 (Drosophila) Hs.436439 -3.449 AMY2A Amylase, α 2A; Pancreatic Hs.484588 -3.438 CCT3 Chaperonin Containing TCP1, Subunit 3 (γ) Hs.491494 -3.433 GSR Glutathione Reductase Hs.271510 -3.421 17-31 10/12/07 14:53 Page 23

INTERNATIONAL JOURNAL OF ONCOLOGY 32: 17-31, 2008 23

Table II. Continued. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Gene symbol Gene name Unigene Score (d) ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– SIAHBP1 Fuse-Binding Protein-Interacting Repressor Hs.521924 -3.415 ACTA2 Actin, α 2, smooth muscle, aorta Hs.500483 -3.413 SNX17 Sorting Nexin 17 Hs.278569 -3.399 ACTG2 Actin, γ 2, Smooth Muscle, Enteric Hs.516105 -3.397 HSP90AB1 Heat Shock Protein 90 kDa α (Cytosolic), Class B Member 1 Hs.649923 -3.389 ZNF43 Zinc Finger Protein 43 Hs.534365 -3.383 PAM Peptidylglycine α-Amidating Monooxygenase Hs.369430 -3.378 EIF3S9 Eukaryotic Translation Initiation factor 3, Subunit 9 η, 116 kDa Hs.371001 -3.368 TUBA3 Tubulin, α 3 Hs.524390 -3.36 RORA RAR-Related Orphan Receptor A Hs.569497 -3.351 TK1 Thymidine Kinase 1, Soluble Hs.515122 -3.349 EIF4E2 Eukaryotic Translation Initiation Factor 4E Member 2 Hs.292026 -3.285 NAT10 N-Acetyltransferase 10 Hs.577281 -3.272 PCGF2 Polycomb Group Ring Finger 2 Hs.371617 -3.259 KIAA0174 KIAA0174 Hs.232194 -3.249 BIRC5 Baculoviral IAP Repeat-Containing 5 (Survivin) Hs.645371 -3.234 SLC35C2 Solute Carrier Family 35, Member C2 Hs.593344 -3.221 EFTUD2 Elongation Factor TU GTP Binding Domain Containing 2 Hs.151787 -3.213 PABPC4 Poly(A) Binding Protein, Cytoplasmic 4 (Inducible form) Hs.169900 -3.2 CSTF2 Cleavage Stimulation Factor, 3' Pre-RNA, Subunit 2, 64 kDa Hs.132370 -3.196 TPI1 Triosephosphate Isomerase 1 Hs.524219 -3.196 ASNA1 ArsA Arsenite Transporter, ATP-Binding, Homolog 1 (Bacterial) Hs.465985 -3.185 GLG1 Golgi Apparatus Protein 1 Hs.201712 -3.182 MLLT7 Myeloid/Lymphoid or Mixed-Lineage Leukemia (Trithorax Homolog, Drosophila); translocated to, 7 Hs.584654 -3.166 DDX39 DEAD (Asp-Glu-Ala-Asp) box polypeptide 39 Hs.311609 -3.165 BAK1 Bcl2-Antagonist/Killer 1 Hs.485139 -3.156 VAT1 Vesicle Amine Transport Protein 1 Homolog (T Californica) Hs.514199 -3.155 EIF3S2 Eukaryotic Translation Initiation Factor 3, subunit 2 ß, 36 kDa Hs.530096 -3.147 P4HB Procollagen-Proline, 2-Oxoglutarate 4-Dioxygenase (Proline 4-Hydroxylase), ß polypeptide Hs.464336 -3.141 LOC729708 LOC729708: Similar to Triosephosphate Isomerase (TIM) (Triose-Phosphate Isomerase) Hs.567941 -3.125 HIST1H3D Histone Cluster 1, H3d Hs.532144 -3.12 PSMB6 Proteasome (Prosome, Macropain) subunit, ß type, 6 Hs.77060 -3.114 EXOC3 Exocyst Complex Component 3 Hs.646923 -3.114 GNL1 Guanine Nucleotide Binding Protein-Like 1 Hs.83147 -3.093 TOMM40 Translocase of Outer Mitochondrial Membrane 40 Homolog (Yeast) Hs.110675 -3.09 MAPKAPK3 Mitogen-Activated Protein Kinase-Activated Protein Kinase 3 Hs.234521 -3.087 RPN1 Ribophorin I Hs.518244 -3.075 SSRP1 Structure Specific Recognition Protein 1 Hs.523680 -3.074 PRNPIP Prion Protein Interacting Protein Hs.132497 -3.062 AP2M1 Adaptor-Related Protein Complex 2, MU 1 Subunit Hs.518460 -3.059 UPP1 Uridine Phosphorylase 1 Hs.488240 -3.059 LY6E Lymphocyte Antigen 6 Complex, locus E Hs.521903 -3.056 ARPC1B Actin Related Protein 2/3 Complex, Subunit 1B, 41 kDa Hs.489284 -3.051 FARSLA Phenylalanine-tRNA Synthetase-Like, α Subunit Hs.23111 -3.044 POLA2 Polymerase (DNA directed), α 2 (70 kDa Subunit) Hs.201897 -3.043 RFC5 Replication Factor C (Activator 1) 5, 36.5 kDa Hs.506989 -3.042 MLF2 Myeloid Leukemia Factor 2 Hs.524214 -3.034 AIP Aryl Hydrocarbon Receptor Interacting Protein Hs.412433 -3.021 MRPL28 Mitochondrial Ribosomal Protein L28 Hs.513230 -3.015 PPP2R4 Protein Phosphatase 2A, Regulatory Subunit B' (PR 53) Hs.400740 -3.008 17-31 10/12/07 14:53 Page 24

24 ZUCCHINI et al: APOPTOTIC GENES AS PREDICTIVE MARKERS IN OSTEOSARCOMA

Table II. Continued. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Gene symbol Gene name Unigene Score (d) ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– KEAP1 Kelch-Like Ech-Associated Protein 1 Hs.465870 -2.985 SNRPN Small Nuclear Ribonucleoprotein Polypeptide N Hs.564847 -2.965 CLIC1 Chloride Intracellular Channel 1 Hs.414565 -2.965 HNRPU Heterogeneous Nuclear Ribonucleoprotein U (Scaffold Attachment Factor A) Hs.166463 -2.958 FTL Ferritin, Light Polypeptide Hs.433670 -2.957 NDUFS2 NADH Dehydrogenase (Ubiquinone) Fe-S Protein 2, 49 kDa (NADH-Coenzyme Q Reductase) Hs.173611 -2.946 RBP1 Retinol binding protein 1, cellular Hs.529571 -2.944 KRT17 Keratin 17 Hs.2785 -2.932 PPP1CA Protein Phosphatase 1, Catalytic Subunit, α isoform Hs.183994 -2.931 HEXA Hexosaminidase A (α polypeptide) Hs.604479 -2.924 HSPA8 Heat Shock 70 kDa Protein 8 Hs.180414 -2.917 TRAP1 TNF Receptor-Associated Protein 1 Hs.30345 -2.904 ATP6V0D1 ATPase, H+ Transporting, Lysosomal 38 kDa, V0 Subunit D1 Hs.106876 -2.901 TNIP1 TNFAIP3 Interacting Protein 1 Hs.543850 -2.901 MARCKSL1 Marcks-Like 1 Hs.75061 -2.892 BCAP31 B-cell Receptor-Associated Protein 31 Hs.522817 -2.854 CRIP2 Cysteine-Rich Protein 2 Hs.534309 -2.854 PRMT1 Protein Arginine Methyltransferase 1 Hs.20521 -2.845 UBTF Upstream Binding Transcription Factor, RNA Polymerase I Hs.89781 -2.837 PSMD13 Proteasome (Prosome, Macropain) 26S Subunit, non-ATPase, 13 Hs.134688 -2.831 KLHDC8B Kelch Domain Containing 8B Hs.13781 -2.806 CCT7 Chaperonin Containing Tcp1, Subunit 7 (ETA) Hs.368149 -2.804 LAMP1 Lysosomal-Associated Membrane Protein 1 Hs.494419 -2.803 HSPA1A Heat Shock 70 kDa Protein 1A Hs.520028 -2.795 TEC TEC Protein Tyrosine Kinase Hs.479670 -2.78 UROD Uroporphyrinogen Decarboxylase Hs.78601 -2.777 SEC13 SEC13-Like 1 (S. Cerevisiae) Hs.166924 -2.768 ZNF496 Zinc Finger Protein 496 Hs.168677 -2.757 FKBP4 FK506 Binding Protein 4, 59 kDa Hs.524183 -2.732 RAB21 RAB21, Member Ras Oncogene Family Hs.524590 -2.727 FADS1 Fatty Acid Desaturase 1 Hs.503546 -2.724 DNAJB1 Dnaj (Hsp40) Homolog, Subfamily B, Member 1 Hs.515210 -2.71 ITPK1 Inositol 1,3,4-Triphosphate 5/6 Kinase Hs.30812 -2.703 DDOST Dolichyl-Diphosphooligosaccharide-Protein Glycosyltransferase Hs.523145 -2.702 UBADC1 Ubiquitin Associated Domain Containing 1 Hs.9194 -2.702 SLC29A1 Solute Carrier Family 29 (Nucleoside Transporters), member 1 Hs.25450 -2.694 KIAA0859 KIAA0859 Hs.647726 -2.692 IGFBP6 Insulin-Like Growth Factor Binding Protein 6 Hs.274313 -2.691 PSMC3 Proteasome (Prosome, Macropain) 26S Subunit, ATPase, 3 Hs.250758 -2.682 KRT5 Keratin 5 (Epidermolysis Bullosa Simplex, Dowling-Meara/Kobner/Weber-Cockayne Types) Hs.433845 -2.682 EIF2S3 Eukaryotic Translation Initiation Factor 2, subunit 3 γ, 52 kDa Hs.539684 -2.653 FOSL2 FOS-Like Antigen 2 Hs.220971 -2.645 CLDN3 Claudin 3 Hs.647023 -2.642 COPZ1 Coatomer Protein Complex, Subunit ζ 1 Hs.505652 -2.636 GUK1 Guanylate Kinase 1 Hs.376933 -2.609 AP2S1 Adaptor-Related Protein Complex 2, σ 1 subunit Hs.119591 -2.585 SMARCE1 Swi/Snf Related, Matrix Associated, Actin Dependent Regulator of Chromatin, Subfamily E, Member 1 Hs.547509 -2.584 FPGS Folylpolyglutamate Synthase Hs.335084 -2.58 HLA-B Major Histocompatibility Complex, Class I, B Hs.77961 -2.578 ID1 Inhibitor Of DNA Binding 1, Dominant Negative Helix-Loop-Helix Protein Hs.504609 -2.576 17-31 10/12/07 14:53 Page 25

INTERNATIONAL JOURNAL OF ONCOLOGY 32: 17-31, 2008 25

Table III. Notable differentially expressed genes involved in biological processes strictly implicated in metastasis. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Gene symbol Gene name Unigene Biological process Mean Log ratio SE ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– VCP Valosin-containing protein Hs.529782 Cell communication/ -1.6052 0.1691 cell death

DHX9 DEAH (Asp-Glu-Ala-His) box polypeptide 9 Hs.191518 Cell death -0.9132 0.1557

BIRC5 Survivin (Baculoviral IAP Repeat-Containing 5) Hs.645371 Cell death -0.5898 0.1485

ARPC1B Actin Related Protein 2/3 Complex, Subunit 1B, 41 kDa Hs.489284 Locomotion -0.9834 0.1716

CTNNAL1 α-catulin [Catenin (cadherin-associated protein) α-like 1] Hs.58488 Cell death/ 0.7292 0.3332 cell adhesion

GADD45A Growth arrest and DNA-damage-inducible, α Hs.80409 Cell growth arrest/cell death 0.734 0.2455

AMFR Autocrine Motility Factor Receptor Hs.295137 Cell communication/ -0.686 0.2492 locomotion

MCAM Melanoma Cell Adhesion Molecule Hs.599039 Cell adhesion -1.0358 0.1748 –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

Results signal transduction, transport, cell death, cell differentiation, cell adhesion, and cell motility. Among all genes selected, The gene expression profiles of both L/B/K ALP- and CD99- great relevance was attached to the genes belonging to two transfected clones were compared to those of the U-2/OS functional categories, namely cell growth arrest/cell death osteosarcoma cell line. Filtering and normalization of raw and cell motility, since these are intimately involved in tumor microarray data were performed using GP3. A total of 2506 progression and invasiveness (Table III). In particular, we cDNAs with GP3 quality-passed values were present in 100% identified several differentially expressed genes involved in of the arrays and were included in the final analysis by the the NF-κB cell survival pathway, a signaling pathway directly SAM program. Four hundred and eighty sequences showed implicated in cancer cell survival and apoptosis, including significantly different levels of expression in both L/B/K ALP- the gene encoding Valosin-containing protein (VCP), DEAH and CD99-transfected osteosarcoma clones, compared to U-2 (Asp-Glu-Ala-His) box polypeptide 9, (DHX9), Survivin OS. Only genes that were up-regulated or down-regulated in (BIRC5) and Growth arrest and DNA-damage-inducible, α at least 85% of the arrays and showing a highly significant (GADD45α). GADD45α, which belongs to a family that change of expression (score of >2.5 or <-2.5) were considered encodes three nuclear proteins related with growth arrest and (Table II). Of these genes, 75 were up-regulated whereas 178 inducible by DNA damage proteins (16,17), is up-regulated were down-regulated in transfected osteosarcoma clones. We in both the transfected osteosarcoma cell lines, showing lack then assigned the different expressed genes to non-mutually of metastatic ability. Since it is known that NF-κB-mediated exclusive functional categories regarding biological processes, repression of GADD45α is sufficient for survival of various using FatiGO. The selected genes covered a broad range of cancer cell types (18), it is conceivable that the up-regulation functional activities, including regulation of cellular growth, of GADD45α, may be dependent on inhibition of the NF-κB 17-31 10/12/07 14:53 Page 26

26 ZUCCHINI et al: APOPTOTIC GENES AS PREDICTIVE MARKERS IN OSTEOSARCOMA

Figure 1. Protein nuclear fraction was extracted and GAPDH staining used to exclude cytoplasmic contamination. As expected GAPDH was evident only in the cytoplasmic but not in the nuclear fraction lysates (A). NF-κB expression was observed in the nuclear fractions of U-2 OS parental cells and transfectants. A lower level of NF-κB expression was observed in the less malignant variants both by Western blotting (B) and by immunofluorescence in adherent cells (C). Digital images were taken in identical conditions, at the same time and using the same image analysis software (Quips-XL genetic workstation, Abbot Visys).

signaling pathway. Indeed, we observed a decrease in the The microarray analysis suggests that inhibition of NF-κB expression levels of NF-κB in the nucleus of U-2OS clones pathway signalling is an important mechanism to restore compared to parental cells (Fig. 1), which reflects a decreased apoptosis/mitosis checkpoints. Quantitative PCR confirmed activity of the nuclear factor. NF-κB is localized mainly in the significant decrease in the expression level of VCP and the cytoplasm in an inactive form bound to an inhibitory Survivin in all of the transfected clones (Fig. 2A). In accordance protein termed IkB (19-21). NF-κB activation occurs via with this, we observed a decrease in the expression of cyclin phosphorylation of IkB, which is essential for the release of D1 (Fig. 2B) and an increase in the apoptotic rate of U-2 OS active NF-κB and its nuclear translocation, where it binds transfected clones (Fig. 3A) that was particularly evident with specific DNA motifs in the promoter regions of target when cells were prevented from adhering (Fig. 3B). This is genes and activates their transcription. In keeping with a interesting since it is well known that non-transformed, reduced NF-κB activity, in less malignant osteosarcoma clones anchorage-dependent cells undergo anoikis (i.e. detachment- we observed down-regulation of both VCP and DHX9, two induced apoptosis) when deprived of contact with the extra- activators of NK-κB pathways, and of Survivin, which is one cellular matrix. In keeping with the lower level of malignancy, of the downstream genes of the NF-κB pathway (Table III). U-2/CD99 and U-2/ALP cell survival was severely reduced 17-31 10/12/07 14:53 Page 27

INTERNATIONAL JOURNAL OF ONCOLOGY 32: 17-31, 2008 27

Figure 2. (A) Relative expression of VCP and Survivin mRNA in U-2 OS parental and derived cells. The relative target gene mRNA expression of U-2 OS parental cells was used as a calibrator (2-ΔΔCT=1). Data are expressed as mean values ± SE. (B) Immunostaining of cyclin D1 on adherent cells. U-2 OS transfectants expressed low levels of cyclin D1 in agreement with their lower malignant potential.

compared to parental cells when cells were cultured in a CD99 and U-2/ALP cells overexpressing α-catulin exhibit condition in which adherence of cells is prevented (15). features of strong adhesion, as characterized by abundant The decrease in Cyclin D1 expression may also be the actin-stress fibers and focal adhesions (Fig. 4B), a pattern consequence of the up-regulation of α-catulin. α-catulin is a which has long been associated with cell differentiation and a member of the vinculin superfamily, whose mRNA expression stationary phenotype. Therefore, besides its antiproliferative appears to be differentially regulated following growth arrest role through decrease in cyclin D1 expression, α-catulin may of cells, suggesting that it may play a role in growth regulation functionally participate in the suppression of transfected U-2OS (22). An antiproliferative role has been in fact demonstrated cell motility (Fig. 5A). From a molecular point of view, the for α-catulin based on attenuation of cyclin D1 transcription dynamic assembly and disassembly of actin filaments and the through inhibition of Ras-mediated signals to the cyclin D1 formation of larger scale filament structures, crucial aspects promoter (23). Thus its overexpression in U-2 OS transfected of actin's function, are under scrupulous control by >100 clones, also confirmed by real-time PCR data (Fig. 4A), is in actin-binding proteins. The actin-related protein-2/3 (ARP2/3) line with the reduction of cyclin D1 expression and the lower complex was the first to be identified of an important set of level of malignancy of transfected clones. In addition, α-catulin actin regulators that initiate formation of new actin filaments can associate with Rho pathway signaling in vivo, serving to (25). Although the precise mechanisms of the ARP2/3 complex localize Rho mediators in a multimolecular signaling complex are still poorly defined, these molecules are involved in the linked to the cytoskeleton (23,24). Acting as a cytoskeletal dynamic actin disassembly during cell migration. In our linker protein, up-regulation of α-catulin in U-2 OS transfected transfected osteosarcoma cell lines there was a down-regulation clones may be associated with the higher level of actin of the ARPC1B gene (Fig. 5B), which encodes p41, a subunit cytoskeleton organization observed in these cells (Fig. 4B), of the human Arp2/3 protein complex. Therefore under- negatively regulating actin disassembly and focal adhesion expression of ARPC1B may contribute to inhibiting disruption dynamics that characterize a motile phenotype. Indeed, U-2/ of actin-containing stress fibers. 17-31 10/12/07 14:53 Page 28

28 ZUCCHINI et al: APOPTOTIC GENES AS PREDICTIVE MARKERS IN OSTEOSARCOMA

253 genes that were up-regulated or down-regulated in at least 85% of the arrays to a highly significant level. These genes are particular intriguing because, being regulated in a similar manner in the majority of transfected clones, they delineate a common transcriptional profile of the metastatic phenotype. These genes can be ascribed to a broad range of functional categories including regulation of cellular growth, signal transduction, transport, cell death, cell differentiation, cell adhesion, and cell motility. Two of the most modulated categories, apart from those related to metabolism, are known to be particularly involved in tumor progression and invasiveness, namely cell growth arrest/cell death and cell motility. Although the molecular mechanisms of metastasis are poorly understood, the resistance to apoptosis seems to be not only one of the required selective advantages that a cell has to acquire in order to form a tumor, but also a crucial process in regulating metastasis. In fact, the induction of programmed cell death is important for the development of the primary tumor because it limits the time available for the accumulation of tumorigenic alterations. However a number of recent experimental observations indicate that crucial apoptotic modulators such as NF-κB, BAX, BCL2, DAPK and survivin are deregulated in metastatic cancer cells (26,27), providing support for the hypothesis that cancer cell survival and programmed cell death escape represent an advantage in metastatic behavior. In our study we have pointed to regulation of the NF-kB cell survival pathway as a major molecular determinant of the metastasis process in osteosarcoma. In particular, we identified several differentially expressed genes involved in the NF-κB cell survival pathway, including GADD45α, VCP, DHX9 and survivin. GADD45α is up-regulated in both transfected osteosarcoma Figure 3. Cytofluorometric analysis of apoptotic U-2 OS cells and derivates cell lines showing a lack of metastatic ability. The GADD45 by annexin-V and propidium iodide after 48 h in standard, monolayer, gene family encodes nuclear proteins (GADD45α, ß, γ) that culture conditions (A) or on polyHEMA-coated dishes (B). Cells (250,000) interact with various cell-cycle-related proteins (16,17). It were seeded on untreated or polyHEMA-coated dishes in IMDM 10% FBS has been demonstrated that NF-κB-mediated repression of and stained for annexin-V after 48 h. GADD45α is essential for various cancer cell types to escape programmed cell death (28). It is conceivable that the reversal of the malignant phenotype observed in our transfected Discussion osteosarcoma cell lines could equally be due in part to the up-regulation of GADD45α, which acts as a tumor suppressor The dissemination of osteosarcoma cells from the primary gene. tumor to the lung is the most frequent cause of death among VCP inhibits apoptosis via degradation of NF-κB inhibitor patients with osteosarcoma. In this study, with a view to (29,30). It has been ascertained that VCP gene expression is identifying genes implicated in the spread of osteosarcoma, associated with the metastatic ability of many human tumors we employed osteosarcoma cell lines showing a reduction in such as hepatocellular carcinoma (31), gastric carcinoma metastatic ability as a result of L/B/K ALP-transfection and (32), pancreatic ductal adenocarcinoma (33) and murine (likewise) CD99-transfection. Our data show that the reversal osteosarcoma cell lines (30). Again, DHX9 is involved as a of the malignant phenotype observed in both differently trans- transcriptional co-activator in NF-κB-mediated gene expression fected clones, may be due mainly to the modified expression and its depletion reduces the NF-κB-dependent transactivation of a set of genes related to growth arrest and apoptosis. (34). In the present study both VCP and DHX9 are down- Recent studies have demonstrated that the forced expression regulated in U-2 OS transfected osteosarcoma cell lines of both CD99 and L/B/K ALP in U-2 OS osteosarcoma cell showing a lack of metastatic ability. In addition, survivin, a lines leads to an identical behavior by the differently transfected member of the inhibitor of apoptosis protein (IAP) family, is cancer cells, that is the abrogation of their invasiveness and likewise down-regulated in our transfected osteosarcoma cell metastatic ability (4,5). By comparing the gene expression lines. Survivin is one of the downstream genes of the NF-κB profiles of transfected clones showing a lack of metastatic pathway (35) and is overexpressed in human cancer cells ability with those of non-transfected osteosarcoma cell lines (36). Furthermore it acts as a regulator of mitosis in the G2-M showing contrasting characteristics, we were able to identify checkpoint representing an important interface between 17-31 10/12/07 14:54 Page 29

INTERNATIONAL JOURNAL OF ONCOLOGY 32: 17-31, 2008 29

Figure 4. (A) Relative expression of CTNNAL1 mRNA in U-2 OS parental and derived cells. The relative target gene mRNA expression of U-2 OS parental cells was used as a calibrator (2-ΔΔC=1). Data are expressed as mean values ± SE. (B) Staining of actin filaments in U-2 OS cells. Cells transfected with CD99 or ALP show a higher level of actin expression and organization in keeping with a less migrating phenotype. Digital images were taken in identical conditions, at the same time and using the same image analysis software (Quips-XL genetic workstation, Abbott Visys).

cell-cycle progression and control of apoptosis (37). Survivin in our transfected osteosarcoma cell lines is particularly overexpression in cancer may obliterate this apoptotic check- intriguing. α-catulin is a member of the vinculin superfamily, point and allow the aberrant progression of transformed cells which includes proteins acting as mediators of cell-extracellular towards mitosis. The decrease in survivin, VCP, and DHX9 matrix adhesion, and seems to be one of the genes related to expression suggests that the observed lack of the malignant anoikis-sensitivity of tumor cells (39). α-catulin has an anti- phenotype in transfected cell lines could be due to the restor- proliferative role based on Ras-mediated inhibition of cyclin ation of apoptotic pathways, and especially sensitivity to D1 promoter and is also involved in the regulation of actin anoikis, which was indeed observed in the clones with reduced polymerization, through its modulation of Rho signaling and malignancy when cells were experimentally prevented from interaction with ß-catenin (23). Thus, α-catulin up-regulation adhering. may explain the decrease in both cell growth and migration Interestingly, Mehlen and Puisieux (38) have convincingly observed in U-2 OS transfected clones. Acting as a cytoskeletal reviewed the role of apoptosis as a multistep barrier that linker protein, α-catulin may participate, together with the modulates the metastasis efficiency at three crucial levels: i) Arp 2/3 complex, in regulation of the dynamic assembly and cell death, by anoikis and amorphosis, after the detachment disassembly of actin filaments and in the formation of larger of primary tumor cells from the extracellular matrix and scale filament structures. α-catulin may favor actin poly- neighboring cells; ii) the death of solitary cells in circulation merization and strong cell adhesion, whereas the Arp 2/3 through tumor immune surveillance or destruction by complex is known to regulate ß-actin polymerization at the mechanical stresses; and iii) cell death during the phase of leading edge and hence cell protrusion, and is thus functionally micrometastasis formation in a secondary organ. In line with connected with a motile phenotype (40). In confirmation of the view that anoikis and amorphosis are crucial mechanisms this pattern, in our transfected osteosarcoma cell lines showing during the metastatic process, the overexpression of α-catulin reduced invasive ability there proves to be an up-regulation 17-31 10/12/07 14:54 Page 30

30 ZUCCHINI et al: APOPTOTIC GENES AS PREDICTIVE MARKERS IN OSTEOSARCOMA

osteosarcoma. Of note, all these genes are functionally related with the NF-κB signaling pathway that appeared to be inhibited in the less malignant osteosarcoma cells, in keeping with a previous study showing that the inhibition of constitutively active NF-κB leads to reversion of malignancy in osteosarcoma (42). Hence, sustained inhibition of NF-κB may be a rational strategy for effective management of this disease.

Acknowledgements

This study was supported by the European Projects EuroBonet (Katia Scotlandi). This work is dedicated to Paolo Carinci, Full Professor of Histology and Embriology at the University of Bologna, who strongly supported this research. Professor Carinci died in July 2007, leaving us the memory of his love for science and life.

References

1. Sporn MB: The war on cancer. Lancet 347: 1377-1381, 1996. 2. Dahlin DC: Bone Tumors: General Aspects and Data on 8,542 Cases. 4th edition. Springfield, IL, 1986. 3. Campanacci M: Bone and Soft Tissue Tumors. 2nd edition. Springer-Verlag, New York, 1999. 4. Manara MC, Baldini N, Serra M, et al: Reversal of malignant phenotype in human osteosarcoma cells transduced with the alkaline phosphatase gene. Bone 26: 215-220, 2000. 5. Manara MC, Bernard G, Lollini PL, et al: CD99 acts as an oncosuppressor in osteosarcoma. Mol Biol Cell 7: 1910-1921, 2006. 6. Banting GS, Pym B, Darling SM and Goodfellow PN: The MIC2 gene product: epitope mapping and structural prediction analysis define an integral membrane protein. Mol Immunol 26: 181-188, 1989. Figure 5. (A) Migration of U-2 OS cells overexpressing CD99. A wound 7. Levy R, Dilley J, Fox RI and Wamk R: A human thymus- healing assay is shown. Images were taken at time 0 and after 8 h from leukemia antigen defined by hybridoma monoclonal antibodies. wounding. (B) Relative expression of ARPCB1 mRNA in U-2 OS parental Proc Natl Acad Sci USA 76: 6552-6556, 1979. and derived cells. The relative target gene mRNA expression of U-2 OS 8. Bernard G, Breittmayer JP, de Matteis M, Trampont P, Hofman P, parental cells was used as a calibrator (2-ΔΔC=1). Data are expressed as mean Senik A and Bernard A: Apoptosis of immature thymocytes values ± SE. mediated by E2/CD99. J Immunol 158: 2543-2550, 1997. 9. Bernard G, Raimondi V, Alberti I, Pourtein M, Widjenes J, Ticchioni M and Bernard A: CD99 (E2) up-regulates alpha4beta1- dependent T cell adhesion to inflamed vascular endothelium under flow conditions. Eur J Immunol 30: 3061-3065, 2000. of α-catulin and down-regulation of ARPC1B. This gene 10. Pettersen RD, Bernard G, Olafsen MK, Pourtein M and Lie SO: encodes p41, one of seven subunits of the human Arp2/3 CD99 signals caspase-independent T cell death. J Immunol 16: protein complex that is involved in assembling and maintaining 4931-4942, 2001. 11. Alberti I, Bernard G, Rouquette-Jazdanian AK, Pelassy C, the structure of the complex itself (41). Underexpression of Pourtein M, Aussel C and Bernard A: CD99 isoforms expression ARPC1B presumably contributes to decreasing dynamic actin dictates T cell functional outcomes. FASEB J 16: 1946-1948, disassembly, a process that lies at the base of cancer cell 2002. 12. Bertaux K, Broux O, Chauveau C, Jeanfils J and Devedjian JC: migration. Moreover one cannot exclude the possibility that Identification of CBFA1-regulated genes on SaOs-2 cells. J in our transfected cell lines a form of cell death might be Bone Miner Metab 23: 114-122, 2005. induced by disruption of the motility processes involving the 13. Fielden MR, Halgren RG, Dere E and Zacharewski TR: GP3: GenePix post-processing program for automated analysis of raw actin cytoskeleton. microarray data. Bioinformatics 18: 771-773, 2002. In conclusion, our study highlights a set of few but 14. Tusher VG, Tibshirani R and Chu G: Significance analysis of significant genes involved in the mitosis progression/apoptosis microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 98: 5116-5121, 2001. switch which may play a pivotal role in the regulation of the 15. Folkman J and Moscona A: Role of cell shape in growth control. metastasis process in the U-2 OS osteosarcoma cell line. It Nature 273: 345-349, 1978. is of great interest that the expression profile of this gene set 16. Hall PA, Kearsey JM, Coates PJ, Norman DG, Warbrick E and Cox LS: Characterisation of the interaction between PCNA and is common to two different experimental models, i.e. two Gadd45. Oncogene 10: 2427-2433, 1995. differently transfected osteosarcoma cell lines (L/B/K ALP 17. Chen IT, Akamatsu M, Smith ML, et al: Characterization of and CD99 clones), showing the same abrogation of metastatic p21Cip1/Waf1 peptide domains required for cyclin E/Cdk2 and PCNA interaction. Oncogene 12: 595-607, 1996. ability. If changes in expression levels of a specific gene set 18. Zerbini LF, Wang Y, Czibere A, et al: NF-kappa B-mediated can be correlated with acquiring or losing metastatic ability, repression of growth arrest- and DNA-damage-inducible proteins such a gene set can be predictive of metastatic outcome. Thus, 45alpha and gamma is essential for cancer cell survival. Proc Natl Acad Sci USA 101: 13618-13623, 2004. despite the fact that the precise involvement of the genes 19. Siebenlist U, Franzoso G and Brown K: Structure, regulation presented here needs to be better elucidated, we propose and function of NF-kappaB. Annu Rev Cell Biol 10: 405-455, this set of genes as a tool to predict the clinical course of 1994. 17-31 10/12/07 14:54 Page 31

INTERNATIONAL JOURNAL OF ONCOLOGY 32: 17-31, 2008 31

20. Shao R, Hu MC, Zhou BP, Lin SY, Chiao PJ, von Lindern RH, 31. Yamamoto S, Tomita Y, Nakamori S, et al: Elevated expression Spohn B and Hung MC: E1A sensitizes cells to tumor necrosis of valosin-containing protein (p97) in hepatocellular carcinoma factor-induced apoptosis through inhibition of IkappaB is correlated with increased incidence of tumor recurrence. J kinases and nuclear factor kappaB activities. J Biol Chem 274: Clin Oncol 21: 447-452, 2003. 21495-21498, 1999. 32. Yamamoto S, Tomita Y, Hoshida Y, et al: Expression level of 21. Simeonidis S, Stauber D, Chen G, Hendrickson WA and valosin-containing protein is strongly associated with progression Thanos D: Mechanisms by which IkappaB proteins control and prognosis of gastric carcinoma. J Clin Oncol 21: 2537-2544, NF-kappaB activity. Proc Natl Acad Sci USA 96: 49-54, 1999. 2003. 22. Zhang J, Nelson M, Wang L, et al: Identification and chromo- 33. Hardy RG, Meltzer SJ and Jankowski JA: ABC of colorectal somal localization of CTNNAL1, a novel protein homologous cancer. Molecular basis for risk factors. BMJ 321: 886-889, 2000. to alpha-catenin. Genomics 54: 149-154, 1998. 34. Tetsuka T, Uranishi H, Sanda T, Asamitsu K, Yang JP, 23. Merdek KD, Nguyen NT and Toksoz D: Distinct activities of Wong-Staal F and Okamoto T: RNA helicase A interacts with the a-catenin family, α-catulin and α-catenin, on ß-catenin- nuclear factor kappaB p65 and functions as a transcriptional mediated signaling. Mol Cell Biol 24: 2410-2422, 2004. coactivator. Eur J Biochem 271: 3741-3751, 2004. 24. Park B, Nguyen NT, Dutt P, et al: Association of Rho guanine 35. Chen X, Kandasamy K and Srivastava RK: Differential roles of nucleotide exchange factor with alpha-catenin-related protein, RelA (p65) and c-Rel subunits of nuclear factor kappa B in tumor alpha-catulin/CTNNAL1, supports serum response factor necrosis factor-related apoptosis inducing ligand signaling. activation. J Biol Chem 277: 45361-45370, 2002. Cancer Res 63: 1059-1066, 2003. 25. Deeks M and Hussey PJ: Arp2/3 and SCAR: plants move to the 36. Span PN, Sweep FC, Wiegerinck ET, Tjan-Heijnen VC, fore. Nat Rev Mol Cell Biol 6: 954-964, 2005. Manders P, Beex LV and de Kok JB: Survivin is an independent 26. Inbal B, Cohen O, Polak-Charcon S, Kopolovic J, Vadai E, prognostic marker for risk stratification of breast cancer patients. Eisenbach L and Kimchi A: DAP kinase links the control of Clin Chem 50: 1986-1993, 2004. apoptosis to metastasis. Nature 390: 180-184, 1997. 37. Li F, Ambrosini G, Chu EY, Plescia J, Tognin S, Marchisio PC 27. Um JH, Kwon JK, Kang CD, et al: Relationship between anti- and Altieri DC: Control of apoptosis and mitotic spindle apoptotic molecules and metastatic potency and the involvement checkpoint by survivin. Nature 396: 580-584, 1998. of DNA-dependent protein kinase in the chemosensitization of 38. Mehlen P and Puisieux A: Metastasis: a question of life or metastatic human cancer cells by epidermal growth factor death. Nat Rev Cancer 6: 449-458, 2006. receptor blockade. J Pharmacol Exp Ther 311: 1062-1070, 39. Kupferman ME, Patel V, Sriuranpong V, et al: Molecular 2004. analysis of anoikis resistance in oral cavity squamous cell 28. Zerbini LF and Libermann TA: Life and death in cancer. carcinoma. Oral Oncol 43: 440-454, 2006. GADD45 alpha and gamma are critical regulators of NF-kappaB 40. Wang W, Goswami S, Lapidus K, et al: Identification and testing mediated escape from programmed cell death. Cell Cycle 4: of a gene expression signature of invasive carcinoma cells 18-20, 2005. within primary mammary tumors. Cancer Res: 64: 8585-8594, 29. Dai RM, Chen E, Longo DL, Gorbea CM and Li CC: Involvement 2004. of valosin-containing protein, an ATPase co-purified with 41. Gournier H, Goley ED, Niederstrasser H, Trinh T and Welch MD: IkappaBalpha and 26 S proteasome, in ubiquitin-proteasome- Reconstitution of human Arp2/3 complex reveals critical roles mediated degradation of IkappaBalpha. J Biol Chem 273: of individual subunits in complex structure and activity. Mol 3562-3573, 1998. Cell 8: 1041-1052, 2001. 30. Asai T, Tomita Y, Nakatsuka S, Hoshida Y, Myoui A, 42. Andela VB, Sheu TJ, Puzas EJ, Schwarz EM, O'Keefe RJ and Yoshikawa H and Aozasa K: VCP (p97) regulates NFkappaB Rosier RN: Malignant reversion of a human osteosarcoma cell signaling pathway, which is important for metastasis of line, Saos-2, by inhibition of NFkappaB. Biochem Biophys Res osteosarcoma cell-line. Jpn J Cancer Res 93: 296-304, 2002. Commun 297: 237-241, 2002.